Compare AI models

Specifications
Parameters
16B1.8B/3.25B
Context window
32k32k
Benchmarks
MATH
27.9%22.8%
MMLU
55.7%55.8%
Overview
CompanyDeepSeekGoogle
Release dateMay 16 2024Dec 6 2023
AccessOpen WeightClosed
Model detailsView modelView model

Frequently asked questions

DeepSeek V2 Lite and Gemini 1.0 Nano are evenly matched across the 2 benchmarks they both report (MATH, MMLU). Gemini 1.0 Nano shipped 162 days before DeepSeek V2 Lite, so benchmark comparisons should account for the intervening progress.

DeepSeek V2 Lite has 16B parameters, while Gemini 1.0 Nano has 1.8B/3.25B. Context windows are 32k (DeepSeek V2 Lite) vs 32k (Gemini 1.0 Nano). DeepSeek V2 Lite is open weight, while Gemini 1.0 Nano is closed.

On MATH, DeepSeek V2 Lite leads at 27.9% vs Gemini 1.0 Nano at 22.8%. On MMLU, Gemini 1.0 Nano leads at 55.8% vs DeepSeek V2 Lite at 55.7%.